Distal Enhancer Potentiates Activin- and GnRH-Induced Transcription of FSHB

Abstract

FSH is critical for fertility. Transcription of FSHB, the gene encoding the beta subunit, is rate-limiting in FSH production and is regulated by both GnRH and activin. Activin signals through SMAD transcription factors. Although the mechanisms and importance of activin signaling in mouse Fshb transcription are well-established, activin regulation of human FSHB is less well understood. We previously reported a novel enhancer of FSHB that contains a fertility-associated single nucleotide polymorphism (rs10031006) and requires a region resembling a full (8 base-pair) SMAD binding element (SBE). Here, we investigated the role of the putative SBE within the enhancer in activin and GnRH regulation of FSHB. In mouse gonadotrope-derived LβT2 cells, the upstream enhancer potentiated activin induction of both the human and mouse FSHB proximal promoters and conferred activin responsiveness to a minimal promoter. Activin induction of the enhancer required the SBE and was blocked by the inhibitory SMAD7, confirming involvement of the classical SMAD signaling pathway. GnRH induction of FSHB was also potentiated by the enhancer and dependent on the SBE, consistent with known activin/GnRH synergy regulating FSHB transcription. In DNA pull-down, the enhancer SBE bound SMAD4, and chromatin immunoprecipitation demonstrated SMAD4 enrichment at the enhancer in native chromatin. Combined activin/GnRH treatment elevated levels of the active transcriptional histone marker, histone 3 lysine 27 acetylation, at the enhancer. Overall, this study indicates that the enhancer is directly targeted by activin signaling and identifies a novel, evolutionarily conserved mechanism by which activin and GnRH can regulate FSHB transcription.

FSH is a key regulator of hormone synthesis and fertility in both sexes. In females, FSH regulates oocyte maturation, preovulatory expression of LH receptor on ovarian granulosa cells, and granulosa cell synthesis of aromatase, an enzyme that catalyzes the conversion of androgens to estrogen (1-3). In males, FSH regulates spermatogenesis and Sertoli cell differentiation and proliferation (4, 5). The importance of FSH in reproduction is underscored by loss-of function (LOF) mutations, which result in infertility in both sexes (6). FSH is a heterodimeric glycoprotein hormone consisting of a unique β-subunit (FSHB) and an α-subunit common to LH, human chorionic gonadotropin, and TSH. Produced within gonadotrope cells in the anterior pituitary, FSH synthesis is rate-limited by transcription of the FSHB gene (7, 8). Women with FSHB LOF mutations fail to initiate puberty and are anovulatory (9-12). Similarly, female Fshb knock-out mutant mice are infertile and anovulatory, with reduced uterine and ovarian size (13). Men with FSHB LOF mutations are azoospermatic, and a subset fail to initiate puberty (11, 14, 15). Although male Fshb knock-out mutant mice are fertile, it is evident that FSH is important for spermatogenesis across species because these mice have reduced sperm counts and motility (13).

During the human menstrual cycle, FSH levels oscillate, with a midcycle peak at the time of the LH surge and a second peak during the late luteal/early follicular phase to initiate follicular maturation (16). Comparable peaks in rodents correspond to the proestrus LH surge and the secondary surge during estrus (17). FSH synthesis and release is driven, in part, by GnRH from the hypothalamus and activins and their inhibitors produced by the ovaries and within the pituitary (18-20).

Activins belong to the TGF-β family of signaling proteins. In gonadotropes, activins bind to their transmembrane type II receptor (likely Activin receptor type 2A) (21), a serine/threonine kinase that recruits and phosphorylates type I receptor partners, Activin receptor type-1B and/or Activin receptor type-1C (22, 23). The type I receptors then phosphorylate SMAD2/3 effector proteins within the cytoplasm, which complex with the common SMAD, SMAD4, and translocate to the nucleus (24). The SMAD complex acts as a transcription factor and recognizes a 4 base-pair (bp) SMAD-binding element (SBE), GTCT, or a higher affinity, 8-bp SBE, which is a palindromic repeat of the 4-bp SBE (GTCTAGAC). SMADs upregulate mouse Fshb transcription in cooperation with Forkhead box protein L2 (FOXL2) (25-28). The combined actions of SMADs and FOXL2 are essential for mouse Fshb expression as conditional deletion of SMAD4 and FOXL2 in gonadotropes results in FSH deficiency and female infertility in a mouse model (29, 30). Activin signaling is opposed by structurally related inhibins, which compete for receptor binding, as well as follistatin, which binds and sequesters activin (18-20, 31). Activin and GnRH have been demonstrated to synergistically induce FSHB transcription (32, 33).

Although activins regulate mouse Fshb through a full, 8-bp SBE at -260/-267 in the Fshb promoter and through several 4-bp SBEs dependent on adjacent FOXL2 binding elements (26-28, 34, 35), the mechanisms of human FSHB regulation by activins remain less clear because the 8-bp SBE is rodent specific. The human -1027/+7 FSHB proximal promoter does contain several FOXL2 binding sites that contribute to activin induction and are adjacent to putative SBEs (26, 36). Even so, the human FSHB promoter is less responsive to activin compared with the rodent Fshb promoter (26, 32, 35, 37). Clinical evidence does support a role for activin in human FSH regulation. Among other observations, FSH deficiency has been observed in individuals with inhibin-secreting tumors (38-41), and postmenopausal women administered an experimental activin inhibitor had reduced FSH levels following treatment (42). For this reason, investigating additional regulatory elements outside the -1028/+7 FSHB promoter could identify additional mechanisms of activin action on FSHB regulation in humans.

We recently discovered a novel enhancer of FSHB located within a highly evolutionarily conserved region spanning ~450 bp and located ~26 Kb upstream of the FSHB transcriptional start site in human (chr11:30,204,683-30,205,132, hg38 assembly) (43). The analogous region from mouse also enhances mouse Fshb transcription and is located ~17 Kb upstream of the Fshb transcriptional start site (chr2:107,076,909-107,077,358, mm10 assembly). Within mouse pituitary, the enhancer was accessible exclusively in gonadotropes and was associated with enrichment of the enhancer marker, histone 3 lysine 4 monomethylation, and the active transcriptional marker, histone 3 lysine 27 acetylation (43, 44).

The human upstream enhancer contains a single nucleotide polymorphism (SNP), which was the most statistically significant variant identified in polycystic ovary syndrome (PCOS) genome-wide association studies, rs11031006 (G/A) (45, 46). In addition to PCOS, the 130-Kb locus at chromosome 11p14.1 containing rs11031006 was also associated with female fertility traits including LH and FSH levels, LH/FSH ratio, testosterone levels, age of natural menopause, and dizygotic twinning (45-52). Compared with the major allele (G), we recently demonstrated that the minor allele (A) increased FSHB transcription in luciferase assays, likely via the creation of a stronger binding site for the transcription factor, Steroidogenic-factor 1/NR5A1 (SF1), a basal regulator of FSHB transcription (43). Besides the SF1 binding site, 2 additional sites within the enhancer were necessary for its activity. One of these sites (256-265) highly resembles an 8-bp SBE (7/8 base pair match).

Based on the presence of the putative SBE, we hypothesized that the enhancer is responsive to activin and that the response would depend on SMAD binding to the identified consensus motif. Consistent with this hypothesis, we found that the upstream enhancer potentiated activin induction of FSHB only when the SBE was intact. Furthermore, SMAD signaling is required for activin potentiation of the enhancer because transfection of the inhibitory SMAD7 reduced enhancer activity. We also determined that the enhancer requires the intact SBE to potentiate GnRH induction of FSHB, consistent with known mechanisms of synergy between activin- and GnRH-responsive elements. In DNA pull-down, SMAD4 bound to an oligonucleotide including the SBE but was not detected when the SBE was mutated. In agreement, SMAD4 was enriched at the enhancer in chromatin immunoprecipitation of gonadotrope-derived LβT2 cells treated with combined activin and GnRH. The combined activin/GnRH treatment also elevated levels of the active transcriptional histone marker, histone 3 lysine 27 acetylation, at the enhancer. Overall, our findings identify a novel role for the FSHB upstream enhancer as a hormone-responsive element and indicate that activin regulation of FSHB transcription may also involve the enhancer as well as the proximal promoter.

Materials and Methods

Plasmids

The human -1028/+7 FSHB luciferase reporter plasmid (-1028/+7 FSHB-luc) in a pGL3 backbone (Promega) was provided by Daniel Bernard (35). The SMAD7 expression vector in a pcDNA3 backbone (Invitrogen) was provided by Aristidis Moustakas (53). The mouse -1000/-1 Fshb promoter luciferase reporter (mFshb-luc), -81/+52 herpes thymidine kinase promoter luciferase reporter (TK-luc), hEnh/G:hFSHB-luc, hEnh/A:hFSHB-luc, hEnh/G:TK-luc, Δ256-265:hFSHB-luc, RVmEnh/G:mFshb-luc, and RVmEnh/A:mFshb-luc were previously described (43). The human hEnh/G:hFSHB-luc and hEnh/A:hFSHB-luc constructs contain a 450-bp enhancer from 26 Kb upstream of the FSHB transcriptional start site (chr11:30,204,683-30,205,132, hg38 assembly) cloned into the KpnI/SacI sites of hFSHB-luc, directly upstream of the -1028/+7 FSHB promoter. At the rs11031006 SNP site, hEnh/G:hFSHB-luc and hEnh/A:hFSHB-luc contain the major (G) and minor (A) alleles of rs11031006, respectively. The hEnh/G:TK-luc construct contains the same human 450-bp major allele enhancer cloned into the KpnI/SacI sites upstream of the TK promoter. Δ256-265 represents a 10-bp deletion within the 450-bp major allele enhancer (chr11:30,204,938-30,204,947, hg38 assembly). The mouse RVmEnh/G:mFshb-luc and RVmEnh/A:mFshb-luc contain a 450-bp enhancer from 17 Kb upstream of the Fshb transcriptional start site (chr2:107,076,909-107,077,358, mm10) cloned into the KpnI/SacI sites of mFshb-luc, directly upstream of the proximal promoter and in the reverse orientation as we had previously determined that it had higher activity than in the forward orientation. At the base equivalent to the human rs11031006 SNP site, mEnh/G:Fshb-luc has the mouse wild-type base (G), equivalent to the human major allele, whereas mEnh/A:Fshb-luc has been mutated to A, equivalent to the human minor allele.

Construction of the enhancer SMAD site mutation plasmid (mutSMAD:hFSHB-luc) was performed using the Quikchange II Site-Directed Mutagenesis Kit (Agilent, cat. #200523) according to the manufacturer’s protocol. The hEnh/G:FSHB-luc plasmid was used as a template. Primer sequences are in Table 1. All plasmids were confirmed by Sanger sequencing (Eton Bioscience).

Table 1.

Primer sequences

Site-directed mutagenesis
Human SMAD mutation	FW:	GGTAGTTTTGTACAGTGGATTATTATAAAAAAAATTTAATAGCACTCTGCTCTTTGATAT
	RV:	ATATCAAAGAGCAGAGTGCTATTAAATTTTTTTTATAATAATCCACTGTACAAAACTACC
qPCR (mouse)
Fshb enhancer	FW:	TGCTCACTGCAAGAAGAGACAG
	RV:	ATAAATATCACAGGGCAGAGCAA
Fshb proximal promoter	FW:	CCCTGTGGATTTACTGGGTGT
	RV:	CGAGGCTTGATCTCCCTGTC
β-actin intron 1	FW:	GGTTTGGACAAAGACCCAGA
	RV:	GCCGTATTAGGTCCATCTTGAG
Ch14 gene desert	FW:	GTCACAGAAACGCAAAGGTTTA
	RV:	CCCAAAGTCATGTTGTACTTGATAG
Hsd17b1 promoter	FW:	CTGTGGGCAGGAGCAGA
	RV:	GCAAGCAAGCGAGCATGAA

Abbreviations: FW, forward; qPCR, quantitative PCR; RV, reverse.

Cell culture and transient transfection

LβT2 cells used in this study were validated by short tandem repeat authentication using the CellCheck19 assay (IDEXX Bioanalytics). The short tandem repeat profile was identical to “LβT2 Cell Stock 2” from a published reference for all 9 of the 4 nucleotide repeat markers (54). Cells were negative for mycoplasma or interspecies contamination.

Cell culture was performed as described (43). Briefly, LβT2 cells cultured in DMEM (Corning, cat #10-013CV) supplemented with 10% fetal bovine serum (Omega Scientific, cat. #FB-01) and 1% penicillin-streptomycin (GE Life Sciences, cat. #SV30010) were maintained at 37°C, 5% CO₂. For luciferase assays, once reaching 80% confluency, LβT2 cells were dissociated by 0.25% trypsin-EDTA (Gibco, cat. #25200056) and plated in 12-well plates at a density of 4.25 × 10⁵/well. The following morning, Polyjet (Sygnagen Laboratories, cat. #SL100688) was used according to the manufacturer’s protocol to cotransfect 500 ng/well of each luciferase reporter plasmid and 200 ng/well of a reporter plasmid encoding β-galactosidase driven by the herpes virus thymidine kinase promoter as an internal control of transfection efficiency. For the experiment including SMAD7, 200 ng/well of SMAD7 expression vector or pcDNA3 empty vector (Invitrogen) was also included. After 5 hours, media was replaced with serum-free DMEM and cells were incubated overnight. For each experimental replicate, three technical replicates were included for each condition.

Hormone treatment and luciferase assay

For activin and GnRH luciferase experiments, approximately 24 hours following transfection, media was removed and replaced with serum-free DMEM containing vehicle (0.001% BSA), 10 ng/μL Activin A (Calbiochem, cat. #114700), and/or 10 nM GnRH (Sigma-Aldrich, cat. #L7134) for a 6-hour treatment. For follistatin treatment, approximately 24 hours following transfection, serum-free DMEM containing either vehicle or follistatin (Novus Biologicals cat # 4889-FN) was added to each well without removal of media. Final concentrations per well were 0.0025% BSA and 250 ng/mL follistatin. Cells were harvested after a 24-hour treatment.

Luciferase assays were performed as previously described (43). Briefly, cells were washed with 1X PBS and agitated for 5 minutes in lysis buffer (0.1 M potassium phosphate [pH 7.8] and 0.2% Triton-X-100) at room temperature. Lysates were split into two 96-well plates. To measure luciferase activity, lysates in the first plate were treated with 4X volume luciferase buffer (25 mM TrisHCl [pH 7.8], 15 mM MgSO4, 10 mM ATP, and 65 μM luciferin). To measure β-galactosidase activity, the second plate was assayed with Galacto-Light Plus reagents (Tropix, cat. #T1009) according to the manufacturer’s protocol. For both assays, luminosity was measured in each well over 1 second following a 1-second delay using a Veritas Microplate Luminometer (Turner BioSystems).

For analysis, luciferase values were normalized to β-galactosidase values from the same tissue culture well (“normalized luciferase values”). Results are expressed as relative luciferase value (RLU), which is defined as the triplicate average of “normalized luciferase values” relative to pGL3 backbone levels from the same hormone treatment group and experimental replicate.

DNA pull-down

DNA precipitation was performed as previously described (35, 55). Following overnight incubation in serum-free DMEM, 10-cm plates of LβT2 cells at approximately 80% confluency were treated with 25 ng/mL Activin A (Calbiochem, cat. #114700) for 1 hour. Protein was collected in 1 mL FLAG buffer (300 mM NaCl, 20 mM Tris [pH 7.5], 1% Triton, 1 mM PMSF [Active Motif, cat. #37495], 1X protease inhibitor [Sigma, cat. #P8340]). Lysates were briefly vortexed, incubated on ice for 15 minutes, and spun at 15 000g for 15 minutes at 4°C before collecting supernatant. The 30 μL streptavidin-coated magnetic beads (Promega, cat. #Z5481) were washed twice in 2X B&W buffer (10 mM Tris [pH 7.5], 1 mM EDTA, 2M NaCl) and incubated with 100 ng/sample biotin-conjugated DNA oligonucleotides for 15 minutes at room temperature in 1X binding buffer (5% glycerol, 20 mM Tris [pH 7.5], 1 mM EDTA, 1 mM dithiothreitol, 0.15% Triton, 100 mM NaCl, 4 mM MgCl). DNA oligonucleotide sequences are listed in Table 2. Beads were washed once in 2X B&W buffer, washed once in 1X binding buffer, and then blocked in 1X binding buffer with 1% BSA for 30 minutes at room temperature. Beads were resuspended in 50 μL 1X binding buffer and incubated for 1 hour at 4°C with cell lysate in the following 500 μL reaction: 100 μL lysate (in FLAG buffer), 150 μL 3X binding buffer, 10 μg Poly DI/DC (Sigma Aldrich, cat # P4929), 50 μL beads. Beads were washed 5 times in 1X binding buffer and eluted by boiling for 5 minutes in 40 μL 1X Laemmli buffer with 200 mM dithiothreitol. The 30 μL/sample was run on a 4% to 20% polyacrylamide gel (Biorad, cat # 4561094), transferred to a polyvinylidene fluoride membrane, blocked for 15 minutes at room temperature in TBS Blocking Buffer (Thermo, cat. #37542), and incubated overnight in SMAD 4 primary antibody (Cell Signaling, cat. #38454) (56) at 1:1000 in TBS blocking buffer with 0.4% Tween. Visualization was performed with goat anti-rabbit IgG-horseradish peroxidase (Santa Cruz, cat. #sc-2004) (57) at 1:10 000 for 1 hour at room temperature. Imaging was done with a Syngene PXi chemiluminescent detector.

Table 2.

DNA pull-down oligonucleotide sequences

mFshb proximal promoter	FW:	Biotin-CAGAAAGAATAGTCTAGACTCTAGAGTCAC
	RV:	GTGACTCTAGAGTCTAGACTATTCTTTCTG
hFSHB enhancer (241-270)	FW:	Biotin-TACAGTGGATTATTATGTCTAGAATTTAAT
	RV:	ATTAAATTCTAGACATAATAATCCACTGTA
SMAD mut hFSHB enhancer (241-270)	FW:	Biotin-TACAGTGGATTATTATAAAAAAAATTTAAT
	RV:	ATTAAATTTTTTTTATAATAATCCACTGTA
mFshb enhancer (248-277)	FW:	Biotin-TACAGCAGATTATTATGTCTGGAATTTAAT
	RV:	ATTAAATTCCAGACATAATAATCTGCTGTA
SMAD mut mFshb enhancer (248-277)	FW:	Biotin-TACAGCAGATTATTATAAAAAAAATTTAAT
	RV:	ATTAAATTTTTTTTATAATAATCTGCTGTA

Abbreviations: FW, forward; RV, reverse.

Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) was performed using the ChIP-IT High Sensitivity kit (Active Motif, cat. #53040). For SMAD4 ChIP, LβT2 cells at approximately 80% confluency in 10-cm tissue culture dishes were treated for 16 hours with 25 ng/mL Activin A (Calbiochem, cat. #114700) and 50 nM GnRH (Sigma-Aldrich, cat. #L7134) in serum-free DMEM. Adherent cells were washed twice with PBS. PBS supplemented with 1 mM MgCl₂ was added to the plate before adding disuccinimidyl glutarate (Thermo Scientific, cat. #20593) in dimethyl sulfoxide to a final concentration of 2 mM. Cells were fixed 45 minutes in disuccinimidyl glutarate, then washed twice with PBS. Cells were fixed for an additional 10 minutes in PBS containing 1% formaldehyde.

For histone H3 lysine 27 acetylation (H3K27Ac) ChIP, LβT2 cells at approximately 80% confluency were treated 16 hours with 25 ng/mL Activin A and 50 nM GnRH or vehicle (0.0075% BSA) in serum-free DMEM. Cells were fixed with Complete Cell Fixation Solution (Active Motif) containing 1.3% formaldehyde for 10 minutes.

For both SMAD4 and H3K27Ac ChIP, chromatin was processed according to the manufacturer’s instructions with a few minor modifications, as previously described (43). Chromatin was sonicated in 300-μL aliquots in 1.5-mL Bioruptor Pico Microtubes (Diagenode, cat. #C30010016) using the Bioruptor Pico (Diagenode, cat. #B01060010), 30 seconds on/30 seconds off for 15 minutes total. A total of 30 μg of chromatin was used per reaction. For SMAD4, 4 μg of SMAD4 antibody (Cell Signaling, cat. #46535) (58) or normal rabbit IgG (Cell Signaling, cat. #2729) (59) were used per reaction. For H3K27Ac, 5 μg of H3K27Ac antibody (Active Motif, cat. #93133) (60) or normal rabbit IgG (Cell Signaling, cat. #2729) (59) were used per reaction. Input was collected from the flow-through of the IgG immunoprecipitation. Input and samples were treated with Proteinase K, purified using the QIAquick PCR Purification kit (Qiagen, cat. #28104), and eluted in a total volume of 100 μL.

Quantitative PCR analysis

DNA from input and immunoprecipitated samples were measured using iQ SYBR Green Supermix (Bio-Rad Laboratories, cat. #1708880) in a CFX Connect Detection System (Bio-Rad Laboratories). A standard curve of serial input dilutions was constructed for each plate and used to compute the concentration of each sample as percent input. A dissociation curve was performed following PCR to ensure the presence of a single product. Three technical replicates were included per experiment. Primer design was previously described (43); sequences are listed in Table 1.

Statistical analysis

For luciferase assays and SMAD4 ChIP, ANOVA followed by post hoc Tukey-Kramer honestly significant difference (Tukey’s HSD) test was performed in Prism 8 (GraphPad). Residuals were checked for normality using the Shapiro-Wilk test with P > 0.05 as the threshold. When needed and as indicated, data were transformed by a Box-Cox transformation in JMP Pro 15 (SAS) or a power transformation and reanalyzed in Prism 8. All transformed data subsequently met the residual normality threshold. P ˂ 0.05 was used as the threshold for statistical significance. For H3K27Ac ChIP, 2-tailed Student t tests adjusted with the Holm-Šídák method for multiple comparison were performed in Prism 8.

Results

The upstream enhancer potentiates activin induction of the FSHB promoter

To determine whether activin treatment increases FSHB transcription driven by the enhancer, human and mouse constructs containing the enhancer cloned upstream of the proximal promoter were transfected into LβT2 cells treated with 10 ng/mL Activin A or vehicle and compared with the promoter alone. At this dose, the human -1028/+7 FSHB promoter alone was not responsive to activin treatment, whereas expression from the enhancer/promoter construct was upregulated 1.7-fold by activin treatment (Fig. 1A). The mouse -1000/-1 Fshb promoter alone was responsive to 10 ng/mL activin (3.9-fold increase compared with untreated promoter), but addition of the mouse enhancer amplified activin induction, with a 7.1-fold increase in response to activin compared with the untreated enhancer/promoter construct (Fig. 1B). For both human and mouse, activin treatment and the enhancer synergistically induced FSHB transcription as determined by 3-way ANOVA (P < 0.05).

Figure 1.

The upstream enhancer potentiates activin and GnRH induction of FSHB promoter. (A) Luciferase expression from reporter constructs driven by the human -1028/+7 FSHB promoter alone (hFSHBp) or downstream of the human major allele enhancer (hEnh/G+p) in LβT2 cells treated with vehicle, 10 ng/mL activin, 10 nM GnRH, or both (n = 9). (B) Luciferase expression from reporter constructs driven by the mouse -1000/-1 Fshb promoter alone (mFshbp) or downstream of the reverse orientation mouse major allele enhancer (RVmEnh/G+p) in LβT2 cells treated with vehicle, 10 ng/mL activin, 10 nM GnRH, or both (n = 9) Each bar represents mean ± standard error of the mean. Data were analyzed by 3-way ANOVA, post-hoc Tukey HSD. Different letters denote significant differences among groups P < 0.05. ‡Indicates a significant 2-way factor interaction (P < 0.05); #indicates a significant 3-way factor interaction. A Box-Cox transform was used on B before analysis. HSD, honestly significant difference; p, promoter; RLU, relative luciferase units.

The enhancer also potentiates GnRH induction of the FSHB promoter and activin/GnRH synergy

Because activin acts synergistically with GnRH to induce transcription from the FSHB proximal promoter (32, 33), we hypothesized that GnRH induction of FSHB transcription would be potentiated by the activin-responsive enhancer. Human and mouse constructs containing the enhancer plus promoter and promoter alone were treated with vehicle or 10 nM GnRH alone or in combination with 10 ng/mL activin. Validating this hypothesis, the human enhancer increased GnRH induction of the human FSHB promoter, with the human promoter-only construct induced 1.4-fold by GnRH treatment compared with 2.0-fold when the enhancer was present (Fig. 1A). Additionally, combined activin/GnRH treatment of the promoter alone increased transcription 1.8-fold, compared with 4.6-fold in the presence of the enhancer. Similarly, the mouse Fshb promoter alone was induced 3.1-fold by GnRH alone and 13.2-fold with combined activin/GnRH treatment, whereas the enhancer/promoter construct was induced 13.3-fold by GnRH alone and 78.5-fold with the combined activin/GnRH treatment (Fig. 1B). For both human and mouse, the synergistic interaction of GnRH and the enhancer and the 3-way enhancer/activin/GnRH interaction were statistically significant by 3-way ANOVA (P < 0.05).

The rs11031006 SNP increases activin induction of mouse Fshb enhancer, but not GnRH induction or induction of the human FSHB enhancer by either hormone

In untreated LβT2 cells, the rs11031006 SNP minor allele (A) increased enhancer activation of basal transcription from the human and mouse FSHB promoters as compared to the major allele (G), likely because of increased binding of the SF1 transcription factor to the enhancer (43). To evaluate whether activin or GnRH treatment would alter this effect, human and mouse constructs containing the minor allele (A) or equivalent mutation in mouse were treated with 10 ng/mL Activin A or 10 nM GnRH (Fig. 2). For the human constructs, although basal transcription was increased by conversion to the minor allele as previously reported, there was no interaction effect on activin induction of the enhancer (Fig. 2A). In contrast, the mouse minor allele equivalent did further potentiate activin signaling (6.3-fold vs 8.0-fold for the major vs minor alleles, respectively) (Fig. 2B). For both species, there was no interaction effect between the minor allele and GnRH on FSHB induction (Fig. 2C and D). Overall, there was no interaction effect between the human rs11031006 minor allele and activin or GnRH induction of the human FSHB promoter, but there was a species-specific interaction between the mouse minor allele equivalent within the enhancer and activin induction of Fshb.

Figure 2.

The rs11031006 SNP has a species-specific effect on activin, but not GnRH, induction of FSHB. (A) Luciferase expression from reporter constructs driven by the human -1028/+7 FSHB promoter downstream of the human major allele enhancer (hEnh/G+p) or minor allele enhancer (hEnh/A+p) in LβT2 cells treated with vehicle or 10 ng/mL activin (n = 4). (B) Luciferase expression from reporter constructs driven by the mouse -1000/-1 Fshb promoter downstream of the mouse major allele enhancer (RVmEnh/G+p) or minor allele enhancer (RVmEnh/A+p) in LβT2 cells treated with vehicle or 10 ng/mL activin (n = 5). (C) Luciferase expression from the human constructs as in panel A, transfected into LβT2 cells treated with vehicle or 10 nM GnRH (n = 4). The same vehicle control groups as in panel A were used for comparison. (D) Luciferase expression from the mouse constructs as in panel B, transfected into LβT2 cells treated with vehicle or 10 nM GnRH (n = 5). The same vehicle control groups as in panel B were used for comparison. Each bar represents mean ± standard error of the mean. Tukey’s HSD was used for post hoc analysis. Different letters denote significant differences among groups P < 0.05. ‡Indicates a significant 2-way ANOVA factor interaction (P < 0.05). A reciprocal cube root transformation was used on panel D before analysis. HSD, honestly significant difference; p, promoter; RLU, relative luciferase units; SNP, single nucleotide polymorphism.

The enhancer is sufficient to mediate activin, but not GnRH, induction of a minimal promoter

Given that the human and mouse enhancers with the major allele similarly potentiated activin and GnRH induction of FSHB (Fig. 1), the human enhancer was used in subsequent experiments aimed at mechanistic interrogation. The FSHB proximal promoter contains several transcription factor-binding sites necessary for basal and hormone-mediated transcription of FSHB, each of which could interact with the enhancer to contribute to transcriptional activity. To determine whether the enhancer alone is truly activin responsive, a minimal herpes virus thymidine kinase promoter (TK) was used in place of the human -1028/+7 FSHB promoter. In vehicle treatment conditions, the enhancer increased transcription relative to the minimal TK promoter alone, consistent with our previous report that the FSHB enhancer is not promoter specific in LβT2 cells (43). Although the minimal promoter alone was not affected by activin treatment, transcription from the human enhancer construct was induced 1.3-fold by activin (Fig. 3A). In contrast, the enhancer was not sufficient to confer GnRH responsiveness (Fig. 3B), indicating that elements within the FSHB proximal promoter are necessary for the increase in GnRH induction mediated by the enhancer.

Figure 3.

The FSHB enhancer confers activin, but not GnRH, responsiveness to a heterologous promoter. (A) Luciferase expression from a reporter driven by the TK minimal promoter alone (TKp) or downstream of the human major allele enhancer (hEnh/G+p) in LβT2 cells treated with vehicle or 10 ng/mL activin (n = 6). (B) Luciferase expression from the same constructs as in panel A, transfected into LβT2 cells treated with vehicle or 10 nM GnRH (n = 6). The same vehicle control groups as in panel A were used for comparison. Each bar represents mean ± standard error of the mean. Data were analyzed by 2-way ANOVA, post hoc Tukey HSD. Different letters denote significant differences among groups P < 0.05. ‡Indicates a significant 2-way ANOVA factor interaction (P < 0.05). HSD, honestly significant difference; p, promoter; RLU, relative luciferase units.

The 256-265 site is necessary for activin and GnRH induction

We hypothesized that activin action on the enhancer is mediated by a previously predicted SBE within 256-265 (Fig. 4A) (43). If so, then deletion of this element within the enhancer construct should reduce transcriptional activation attributed to activin treatment. To investigate whether the SBE contributes to activin induction of FSHB, LβT2 cells transfected with constructs containing a deletion of 256-265 within the enhancer cloned upstream of the proximal FSHB promoter (hΔ256-265hEnh+p) were treated with 10 ng/μL Activin A (Fig. 4B). As previously demonstrated (43), basal transcriptional activity of the Δ256-265 enhancer was reduced to promoter-only levels. Furthermore, there was no response to activin treatment, indicating that the 256-265 site is necessary for both basal and activin-stimulated transcriptional activity regulated by the enhancer (Fig. 4C). Because a deletion could also reduce transcriptional activation by creating a new binding site for a repressor or by changing the spacing between other regulatory elements, a specifically mutated SMAD site (human SBE mutation) construct was also studied (Fig. 4A). Basal transcriptional activity of the mutant was reduced compared to the wild-type enhancer, although it was still elevated above promoter-only levels (Fig. 4D). Comparable to the deletion construct, transcription was not increased in response to activin treatment (Fig. 4E).

Figure 4.

The 256-265 site is required for maximal activin and GnRH induction via the FSHB enhancer. (A) Sequence comparison of the SMAD consensus with the putative SBE within the human enhancer at 256-265, the equivalent site within the mouse enhancer at 263-272, and the human SBE mutation construct. Bases mismatched with the SMAD consensus sequence are bolded. (B-I) Luciferase expression from transfection of LβT2 cells with reporter constructs driven by the human -1028/+7 FSHB promoter alone (hFSHBp), downstream of the full human major allele enhancer (hEnh/G+p), or downstream of the human major allele enhancer with deletion (hΔ256-265+p) or mutation (hSBEmut+p) of the putative SBE and treatment as indicated. (B) SBE deletion construct (hΔ256-265+p) with vehicle or 10 ng/mL activin treatment (n = 6). (C) Data from panel B expressed as fold induction (activin-treated/vehicle). (D) SBE mutation construct (hSBEmut+p) with vehicle or 10 ng/mL activin treatment (n = 5). (E) Data from panel D expressed as fold induction (activin-treated/vehicle). (F) SBE deletion construct (hΔ256-265+p) with vehicle or 10 nM GnRH treatment (n = 6). The same vehicle control groups as in panel B were used for comparison. (G) Data from panel F expressed as fold induction (GnRH-treated/vehicle). (H) SBE mutation construct (hSBEmut+p) with vehicle or 10 nM GnRH treatment (n = 5). The same vehicle control groups as in panel D were used for comparison. (I) Data from panel H expressed as fold induction (GnRH-treated/vehicle). Each bar represents mean ± standard error of the mean. Data were analyzed by (B, D, F, and H) 2-way or (C, E, G, and I) 1-way ANOVA, post hoc Tukey HSD. Different letters denote significant differences among groups P < 0.05. A Box-Cox transform was used on panels F and H before analysis. p, promoter; RLU, relative luciferase units; SBE, SMAD-binding element.

Because the enhancer also potentiated GnRH induction of FSHB but was not sufficient to mediate GnRH induction of a minimal promoter (Figs. 1A and 3B), we hypothesized that GnRH potentiation mediated by the enhancer may require synergy with activin-responsive elements. For this reason, we investigated whether deletion or mutation of the SBE within the FSHB enhancer would affect GnRH potentiation. Compared with the intact enhancer construct, deletion of the SMAD site modestly reduced transcriptional activation by GnRH (Fig. 4F), although it was still elevated above promoter-only levels (Fig. 4G). Results obtained from the SMAD mutation construct were comparable (Fig. 4H and I). Overall, the putative SBE within 256-265 contributes to transcriptional activation of the enhancer in response to both activin and GnRH treatment, although the overall effect of the enhancer and SBE on GnRH induction is weaker as compared to activin.

Follistatin represses enhancer-mediated FSHB transcription

Deletion of the putative SBE 256-265 within the enhancer was sufficient to reduce transcription to promoter-only levels (Fig. 4B), suggesting that activin activation and subsequent binding of SMADs is necessary for enhancer activity. Because LβT2 cells synthesize and secrete activin (61), untreated cells are continuously exposed to activin signaling, limiting the evaluation of the necessity of activin signaling for enhancer activity. To circumvent this limitation, transcriptional activity of the human enhancer was measured in the presence of the activin repressor, follistatin. Follistatin represses activin signaling by sequestering activin, preventing it from associating with its cell-surface receptor (31). Follistatin treatment had no effect on transcription from the FSHB promoter alone but reduced activity from the human enhancer construct to promoter-only levels (Fig. 5A), demonstrating that activin signaling is necessary for enhancer activity.

Figure 5.

The classical SMAD signaling pathway is required for activin induction of the FSHB enhancer. (A) Luciferase expression from reporter constructs driven by the human -1028/+7 FSHB promoter alone (hFSHBp) or downstream of the major allele enhancer (hEnh/G+p) in LβT2 cells treated with vehicle or 250 ng/mL follistatin (n = 5). ‡Indicates a significant 2-way ANOVA factor interaction (P < 0.05). (B) Luciferase expression from reporter constructs as in panel A, cotransfected with 200 ng/well of a SMAD7 expression vector or empty vector in LβT2 cells. Each bar represents mean ± standard error of the mean. Data were analyzed by (A) 2-way or (B) 3-way ANOVA, post hoc Tukey HSD. Different letters denote significant differences among groups P < 0.05. HSD, honestly significant difference; p, promoter; RLU, relative luciferase units.

Classical SMAD signaling is required for activin induction of the FSHB enhancer

We next sought to determine whether the classical activin signaling pathway mediated by phosphorylation of SMAD transcription factors is required, as would be expected if activin induction of the enhancer results from SMAD binding. SMAD7 acts as a dominant-negative inhibitor of SMAD phosphorylation by binding to the TGF-β type I receptor, blocking its ability to associate with and phosphorylate other SMAD proteins (62). Although SMAD7 overexpression had no effect on transcription from the FSHB promoter alone, it reduced both basal and activin-treated transcriptional activity from the enhancer to 0.8- and 0.6-fold, respectively, compared with transfection of the empty expression vector (Fig. 5B), indicating that the classical SMAD signaling pathway is necessary for basal and activin-regulated enhancer activity.

SMAD4 binds to the enhancer

To identify whether SMADs could bind to the putative SBE at 256-265, DNA pull-down was performed using biotin-labelled, 30-bp DNA double-stranded oligonucleotides spanning 241-270 of the human FSHB enhancer to capture proteins from whole cell lysates of LβT2 cells treated for 1 hour with 25 ng/mL Activin A (Fig. 6A). Although SMAD4 was detected by an oligonucleotide representing the endogenous sequence, it was not captured when the putative SBE was mutated (same mutation as indicated in Fig. 4A). Similar results were obtained with oligonucleotides spanning 248-277 of the mouse Fshb enhancer to capture the analogous region.

Figure 6.

SMAD4 binds to the FSHB enhancer. (A) Biotin-labelled 30 bp double-stranded oligonucleotides from the indicated regions of the human and mouse FSHB enhancers containing the wild-type (WT) or mutant SBE (SBE mut) were used in DNA pull-down of protein from activin-treated LβT2 whole cell lysates. Following precipitation of DNA/protein complexes with streptavidin magnetic beads, proteins were eluted and analyzed by Western blot with a SMAD4 antibody. Input is shown for comparison. Image is representative of four replicates. (B) ChIP-qPCR using a SMAD4 antibody (S4) and normal rabbit IgG control to compare binding at the negative control region (chromosome 14 gene desert), positive control region (Hsd17b1 promoter), Fshb enhancer, and Fshb proximal promoter in LβT2 cells treated overnight with 25 ng/mL Activin A and 50 nM GnRH (n = 3). Data were analyzed by 2-way ANOVA, post hoc Tukey HSD. Different letters denote significant differences among groups P < 0.05. Ch14, chromosome 14; Enh, enhancer; HSD, honestly significant difference; p, promoter; qPCR-ChIP, quantitative PCR-chromatin immunoprecipitation; S4, SMAD4 antibody.

To verify that SMADs can bind to the upstream enhancer in native chromatin, we performed ChIP using an antibody against the common SMAD, SMAD4 (Fig. 6B). As compared with the negative control region (chromosome 14 gene desert), SMAD4 binding was enriched at the Fshb enhancer and positive control Hsd17b1 promoter (63), although this assay was not sensitive enough to detect it at the Fshb proximal promoter.

Activin and GnRH treatment elevates H3K27Ac at the enhancer

The histone marker H3K27Ac is associated with active transcription (44). To determine whether activin and GnRH treatment could increase H3K27Ac enrichment near the Fshb enhancer, ChIP was performed on LβT2 cells treated with a combination of 25 ng/mL Activin A and 50 nM GnRH and compared with vehicle (Fig. 7). although the Fshb proximal promoter was unaffected by the treatment, H3K27Ac at the Fshb enhancer was increased 2-fold compared with vehicle. These results support a role for activin and GnRH in increasing enhancer activity.

Figure 7.

H3K27Ac increases at the Fshb enhancer following activin and GnRH treatment. ChIP-qPCR was performed using an H3K27Ac antibody to compare enrichment at the negative control (chromosome 14 gene desert), positive control (b-actin intron 1), Fshb enhancer, and Fshb proximal promoter regions between activin/GnRH and vehicle treatment groups in LβT2 cells. Enrichment using a normal rabbit IgG control is shown for each treatment and region for comparison (n = 3). At each locus, the vehicle-treated group served as a control for the GnRH/activin-treated group and was compared by 2-tailed Student t test. P values are adjusted with the Holm-Šídák method for multiple comparisons. *P < 0.05. Ch14, chromosome 14; Enh, enhancer; p, promoter; qPCR-ChIP, quantitative PCR-chromatin immunoprecipitation.

Discussion

Although activin has long been recognized as an activator of human FSHB transcription, the mechanism has remained somewhat elusive given the lack of conservation of the rodent 8-bp SBE within the proximal promoter and the relatively weaker induction of the human FSHB proximal promoter by activin (26, 32, 35, 37). We identified a role for a recently discovered upstream enhancer of FSHB in activin and GnRH regulation of FSHB and determined that an imperfect, full SMAD site within the enhancer is necessary for its function. The ability of the upstream element to regulate activin- and GnRH-stimulated FSHB transcription is conserved between human and mouse. Through DNA pull-down and ChIP experiments, we confirmed that SMAD4 can bind to this element. Furthermore, combined activin and GnRH treatment increased H3K27Ac recruitment at the upstream enhancer, supporting the role of the enhancer as a hormone-responsive regulator of FSHB transcription.

The FSHB enhancer described here likely acts in cooperation with the FSHB promoter and other regulatory elements. A randomly integrated, 10-Kb human transgene encoding FSHB flanked by 4 Kb of upstream and 2 Kb of downstream sequence was previously demonstrated to restore fertility to an Fshb knock-out mouse but could not compensate for a loss of FOXL2 and SMAD4 function, suggesting that 1 or both factors act on the 10-Kb transgene to regulate human FSHB expression in this mouse model (64, 65). Interestingly, activin induction of the human transgene was weaker than the mouse gene in primary pituitary cultures, potentially reflecting that additional elements not included in the transgene are needed for full activin induction. Our results confirm the presence of an additional activin-responsive element within the upstream enhancer that was not included within the 10-Kb transgene and support a mechanism that includes direct regulation by SMAD binding to the enhancer.

On the human FSHB proximal promoter, activin action is dependent on FOXL2 binding sites located immediately adjacent to SMAD half-site binding elements. In a search using the Find Individual Motif Occurrences tool (66) with the JASPAR (67) FOXL2 binding matrix (MA1607.1) as input and P < 10^-4 as the cutoff for match discovery, no putative binding sites were identified within the human enhancer, although the more general Forkhead class (MF0005.1) binding matrix did identify 1 match within the human enhancer from 288-296. Because deletion of this element did not affect activity of the enhancer in basal conditions but deletion of the SMAD element at 256-265 within the enhancer did (43), we focused our investigation on the latter. Given that the 256-265 enhancer SMAD site contains only a single mismatch compared with the full 8-bp canonical SBE, it is indeed possible that FOXL2 is not required for SMAD binding at this locus. Similarly, the full SMAD site within the mouse proximal promoter is not immediately adjacent to a FOXL2 element but contributes to SMAD induction of Fshb (34, 35, 37). Additionally, the SMAD site on the human enhancer is adjacent to another element required for basal enhancer activity that is a perfect match to the consensus site for the transcription factor PTX1, which has been shown to directly interact with SMADs in gonadotropes (68, 69).

Direct regulation of the enhancer by activin is supported by the necessity of the intact SBE within the enhancer, detection of SMAD4 binding to the enhancer in ChIP and DNA pull-down, and sufficiency of the enhancer to confer activin responsiveness to a minimal promoter. In contrast, GnRH potentiation mediated by the enhancer requires the FSHB proximal promoter, which has previously been reported to be strongly responsive to GnRH in both species even in the absence of the enhancer (32, 33), although we did observe variability in the strength of GnRH induction of the human FSHB promoter in our experiments (compare Fig. 1A with Figs. 4F, H). Furthermore, combined activin/GnRH synergy was also potentiated by the enhancer, and GnRH induction was dependent on the SMAD full site. We therefore conclude that GnRH is most likely acting at the proximal promoter and synergizing with activin-responsive elements on the enhancer, although we have not formally ruled out the possibility that GnRH effector proteins (such as FOS/JUN) could directly bind to the enhancer.

We had previously reported that the active transcriptional marker H3K27Ac was enriched at the enhancer in chromatin from whole pituitary, but we were unable to detect significant enrichment in untreated LβT2 cells. We hypothesized that this difference may be due to hormone action required for activation of the enhancer. In this study, H3K27Ac was detectable in LβT2 chromatin at the enhancer even with vehicle treatment. This difference may be due to technical differences between the 2 experiments, such as media composition at the time of fixation or sonication duration or to variability intrinsic to the LβT2 cell line (54). Regardless of this difference, our current findings do reveal that activin and GnRH treatment results in accumulation of H3K27Ac, supporting our hypothesis that these factors activate the enhancer.

Given that the direction of the rs11031006 effect on basal FSHB transcription was counter to our initial hypothesis based on FSH levels in PCOS patients (increasing rather than decreasing FSHB transcription), we had previously speculated that the rs11031006 SNP might interfere with activin or GnRH induction of the enhancer. This does not seem to be the case. In mouse, a mutation analogous to the human rs11031006 minor allele synergistically increased activin induction, although this effect was not observed with the human constructs, potentially reflective of species-specific differences between the human and mouse SF1 and SMAD elements. Although the human major allele SF1 site containing rs11031006 matched 7/8 bp to the SF1 consensus, the mouse sequence contains an additional mismatch, resulting in a potentially greater increase in SF1 binding with conversion to the minor allele equivalent (43). Similarly, the mouse SMAD site also contains an additional mismatch with the SMAD consensus. Together, SMAD binding to the mouse Fshb enhancer may be more dependent on anchoring by other factors, and the stability of a complex containing both SF1 and SMAD could be more dramatically increased by conversion to the rs11031006 minor allele equivalent. Although not synergistic, the effect of the rs11031006 minor allele from human was additive with activin and GnRH signaling, suggesting that it does still increase FSHB transcription above major allele levels as a result of increased basal transcription. Whether this effect could affect human fertility remains to be investigated.

In summary, we determined that the FSHB enhancer potentiates activin and GnRH induction of FSHB transcription, dependent on a SBE at 256-265. With respect to both basal and activin/GnRH regulation, the function of the FSHB enhancer is conserved between human and mouse, presenting an opportunity for future investigations of the enhancer and rs11031006 SNP with mouse models.

Our findings contribute to a greater understanding of the mechanisms of FSHB regulation. FSH plays an indispensable role in human reproduction through regulation of puberty and fertility (9-15), but also contributes to nonreproductive functions, such as bone density and adiposity (70-72). Although FSH insufficiency is associated with anovulation and conditions such as PCOS, elevated FSH levels can contribute to bone loss in postmenopausal women and premature ovarian failure (71, 73); therefore, both extremes are consequential for human health. Because FSH synthesis is rate-limited by FSHB transcription (7, 8), understanding how the FSHB gene is regulated may provide valuable insight for novel therapeutic targets.

Glossary

Abbreviations

bp

base pair

ChIP

chromatin immunoprecipitation

HSD

honestly significant difference

LOF

loss-of function

PCOS

polycystic ovary syndrome

RLU

relative luciferase value

SBE

SMAD binding element

SF1

Steroidogenic-factor 1/NR5A1

TK-luc

thymidine kinase promoter luciferase reporter

Additional Information

Disclosures: Authors have no conflict of interest.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.